This paper investigates the cues triggering an echo or an information seeking interpretation from string identical utterances. It provides evidence that prosody is not accurate and reliable in predicting the interpretation of the inquiry as echo or information seeking. Based on these results we propose to model the differences between EchoQ and InfoS by appealing to where in the context update the utterance is located. Our analysis allows us to have a unified semantics for EchoQs and Infos while keeping the interpretation true to form.